In this paper a new authentication system using Finger Knuckle Surface is examined. This introduces a personal authentication system that can simultaneously extract and exploit Finger back Knuckle surface geometrical features. Unlike, existing work on hand and finger geometrical methods which mainly concentrates of features extraction and recognition, this methods experiments with subsets of extracted feature to achieve better performance by exploiting less number of features. This is achieved by the determination of hybrid convex curves from the finger back knuckle surface. From the identified feature curves, the subset of the features like Knuckle edge points and knuckle tip points were identified. From these identified contours, geometrical structures like tangents and secants were constructed to obtain feature information in terms of angle. This method reduces critical problems that arise due to the extraction of more number of features. Also reduces the computational complexity of the feature extraction and recognition process. © 2013 IEEE.